You are here

How to determine the value size in block of *.cst files?

2 posts / 0 new
Last post
How to determine the value size in block of *.cst files?

HELLO, everyone,

I am a begineer of Rosetta de nevo peotein design and want to design an enzyme using rosettamatch, however it is complicated even i have read its method carefully. Recently, I am trying to use demos, and i found that are really helpful and important. But some used files need to be made by hand, such as the *.cst file. The question is how to determine the value size in block of *.cst files to constrain a ligand and residue interaction, including :

The 2nd clumon, xtol, specifying the allowed tolerance xtol of the value. The 3rd column, specifying the force constant k, or the strength of this particular parameter; The 4th column, the periodicity per of the constraint; And, 5TH the number of sampling points between x0 + xtol and x0 - xtol.

I want to know is there such kind of video for Rosettamatch. Could anyone please sent a link or example? Thanks sincerely.


Wei Fulei,

from East China University of Science and Technology, China.

Post Situation: 
Sun, 2019-05-05 19:56

I don't believe there's a video tutorial about RosettaMatch.

Regarding the values in the block, it's something of a judgement call. I'd be guided by similar values as what someone else has successfully used for their paper, but there's likely to be some changes based on your particular system and desires.

The second column, xtol, specifies how precise you need to be around the given value. This likely will be guided by the type of interaction you're doing and how critical it is in your system to be close to the center value you specify. A good heuristic for figuring it out is to look at comparable native interactions, and look at the acceptable range of values for comparable interactions. (e.g. look at the natural variation of certain hydrogen bonding interactions). You may then want to increase or decrease that value depending on how critical you think matching the geometry is to your system.

The third column, k, is only used during the design portion of the protocol, not the match portion. This tells Rosetta how much to penalize values which fall outside of the range specified by xtol. The bigger the value, the more it's penalized. Again, what you set this to depends a bit on how "hard" you want to make the window, and if you're okay with slight violations. (E.g. how frequently similar interactions go outside the window.) Often it comes down to guess-and-check. If you aren't getting the designs you want, you might want to try softening one or more of these values to allow them to go outside the window. (Generally, though, I'd start with similar values as used successfully in other papers.)

The fourth column is the periodicity. This is typically only changed for dihedral angles(*), and left at 360 for regular angles. The periodicity allows you to do things like say that you want the interaction to be planar, but you don't care if it's cis or trans (set the periodicity to 180, making the x0 = 180 value mean both 0 and 180 degrees) or you can specify that you want anti/gauche+/gauche- by setting periodicity to 120, making the x0 = 180 value mean 180, 60 and 300.

*) The periodicity column has a special meaning for distance line. It should be 0 or 1, with a "1" indicating that the distance being represented is a covalent bond. This will change things internally in Rosetta, actually annotating things as a bond in the Pose, turning off the atom-atom repulsive interaction between the atoms. (This would otherwise be on for a non-covalent interaction.

The fifth column is only used in the match stage itself (not in the design portion). This tells the matcher how many samples to use in the xtol region around x0. The greater the number, the more samples, and the more chance of getting a match due to a precise alignment of geometries, but the more samples you go through the longer matching will take, the more memory it will use, and the more results you're likely to get. Generally I'd recommend starting with a small value, and only increase it if you're not getting enough match results.

Tue, 2019-07-09 15:48